segmentationfundraisinghow-to

How to Segment Your Donor Database Without a Data Analyst

BH
Brian Ho
||4 min read

Most guides to donor segmentation assume you have someone on staff who can build a pivot table, or better yet, a dedicated data analyst. Most small and mid-size Canadian charities have neither. This is a segmentation approach that works with what a two-person fundraising team actually has: a donor list, a spreadsheet, and an afternoon.

Start with three questions, not twelve segments

Sophisticated donor segmentation can involve a dozen overlapping categories. For a team without a dedicated analyst, that's a plan that looks good on paper and never gets built. Start instead with three questions about every donor:

  • When did they last give? (recency)
  • How often do they give? (frequency)
  • Is their giving growing, flat, or shrinking? (trend)

Those three questions, applied consistently, produce a working segmentation without any statistical model at all — and they're the same three signals a formal propensity model starts from, just applied by hand instead of automatically.

A segmentation you can build in a spreadsheet this afternoon

If your donor list is a few hundred records or fewer, this is realistic to do directly in Excel or Google Sheets:

  • Warm: gave within the last 90 days
  • Cooling: gave 91 to 365 days ago
  • At risk: gave 12 to 24 months ago
  • Lapsed: no gift in over 24 months

Add a second pass for trend: within each group, flag anyone whose most recent gift was larger than their average of the past three gifts as an "upgrade candidate," and anyone whose most recent gift was smaller as a "watch" — a soft signal that something may be changing before it shows up as a full lapse.

This gets you four groups and one cross-cutting flag, which is enough to send meaningfully different messages to meaningfully different donors — without a single formula more complex than a date difference and an average.

What this looks like with real (illustrative) numbers

To make the four groups concrete, here's how five sample donors would sort using just the recency/frequency/trend rules above. None of these are real donor records — they're illustrative only.

| Donor | Last gift | Typical frequency | Recent trend | Segment | |-------|-----------|-------------------|--------------|---------| | A. Chen | 45 days ago | Every ~60 days | Flat | Warm | | M. Boucher | 220 days ago | Every ~120 days | Growing (last gift +30%) | Cooling — upgrade candidate | | R. Okafor | 540 days ago | Every ~90 days | N/A — overdue | At risk | | T. Nguyen | 3 years ago | Was every ~180 days | N/A — inactive | Lapsed | | S. Patel | 20 days ago | Every ~30 days | Shrinking (last gift -25%) | Warm — watch flag |

Notice that M. Boucher and S. Patel would both look like ordinary, unremarkable donors on a simple recency report alone — one is 220 days out, the other gave three weeks ago. It's the trend column that flags them: one as a quiet upgrade opportunity, the other as an early watch signal, worth a check-in before a full year passes and they become a "cooling" or "at risk" donor with no explanation why.

Where manual segmentation starts to break down

This approach works well up to a few hundred donors. Past that, three things tend to happen: the spreadsheet becomes too large to sanity-check by eye, the segments stop being refreshed regularly because updating them by hand takes real time each cycle, and — most importantly — the segmentation stops accounting for anything beyond recency and frequency, like which specific campaigns a donor has responded to, or how their personal giving cycle compares to the charity's overall pattern rather than a flat 90/365-day rule that doesn't fit every donor equally.

None of that means manual segmentation was the wrong place to start. It means it's the floor, not the ceiling — and knowing where that ceiling is helps a fundraising team decide when it's actually worth automating, instead of guessing.

What automated segmentation adds

A propensity-based tool doesn't replace the four groups above — it refines them. Instead of a flat 90-day cutoff for every donor, it can compare each donor's recency against their own historical giving cycle (a donor who gives every 45 days and is at day 60 is a different priority than one who gives every 400 days and is at day 60, even though both technically fall in the same flat time window). It can also fold in signals a spreadsheet realistically can't track by hand at scale: cause affinity, campaign-specific response history, and — for Canadian charities specifically — CASL consent status per donor, per channel.

The practical difference shows up as a ranked list with a plain-language reason attached to each name, rather than a static bucket a donor sits in until someone remembers to update the sheet.

A realistic path from manual to automated

If you're starting from nothing, the manual four-group segmentation above is a genuinely good first step — it costs nothing but time, and it will immediately improve on a single undifferentiated broadcast list. Treat it as validation: if splitting your list into warm/cooling/at-risk/lapsed and sending different messages to each produces a noticeably better response than last year's blanket appeal, that's a strong signal a more granular, automatically-refreshed version of the same idea is worth the next step.

Ready to move past the spreadsheet?

GivingSignal's Donation Likelihood tool runs this same recency/frequency/trend logic automatically against your donor CSV, adds cause affinity and CASL consent flags, and refreshes every time you upload new data — no formulas to maintain. And the free CASL Consent Tracker is a good companion first step, even before you run a full scoring tool.

BH
Brian Ho
Contributing Writer at GivingSignal

More from the blog

Your donors are ready

Upload your donor CSV and get propensity-scored insights in minutes. Built for Canadian charities, CASL-compliant by design.

Start your free 14-day trialLearn more